subscribe

NRF 2017 – 2 AI and Analytics in Retail

One of the pre-show press releases from the NRF said, “New innovations and technical capabilities will continue to transform the retail experience, from mobile to personalization to bringing artificial intelligence and augmented reality out of the future and into the present.”

Artificial Intelligence (AI) was certainly at the show, including the IBM booth that highlighted the application of the Watson AI software to retail problems. IBM said 1-800-Flowers and Staples were two IBM partners using Watson to help with their operations. Frankly, IBM reps and the displays at the NRF didn’t make it clear how Watson’s AI differed from “Big Data” analytics in common use throughout retail.

Besides IBM, other big data companies such as Oracle were at the NRF show, along with a host of smaller software companies that specialize in the retail market.

Analytics does not just describe after-the-fact analysis of sales, although that is an important part of it. It can also be used in real time to keep track of who is in the store and what they are buying, plus what they are looking at but not buying. The systems are often coupled with machine vision systems that not only counted people but broke them down demographically. These machine vision systems are claimed to be able to identify gender, race or ethnicity and rough age from the camera images. They could also identify people and track individuals as they moved through the store. The system demonstrated at the Google booth could, it was said, even determine the emotions of the individuals. These demonstrations were not convincing and had very high error rates, even on simpler things like gender. The Google software was simply unable to perform except for extremely exaggerated facial expressions.

Google Facial Recognition 01 resizeGoogle claimed its facial recognition system could detect four emotions and color code them: yellow for joy (happiness), red for anger, blue for sorrow and purple for surprise. While this may have worked for a specially trained Google employee showing extreme and stereotyped emotions, it didn’t work for me or for many other visitors to the Google booth.

Analyst Comment

Clearly, Matt did not see a good demonstration of analytics, but data shown at the Munich Digital Signage event last Autumn seemed to show that the analytics are really useful. (Vizualize Can Track Individuals Without Privacy Problems.) BR